Applied Energy 205 (2017) 33–43

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

A method for assessing infrastructure for CO2 utilization: A case study of MARK
Finland
⁎
Hannu Karjunen , Tero Tynjälä, Timo Hyppänen
Lappeenranta University of Technology, Skinnarilankatu 34, PL 20, 53851 Lappeenranta, Finland

H I G H L I G H T S

• Modeling tool is presented for comparing carbon capture and utilization strategies.
• The temporal disparity between CO production and consumption is identified.
2

• Key decisions defining a general carbon utilization system are proposed.

A R T I C L E I N F O A B S T R A C T

Keywords: Synthetic hydrocarbons can be produced sustainably with power-to-gas processes, resulting in a net reduction of
Carbon dioxide greenhouse gas emissions due to the substitution of conventional natural gas and other fossil fuels with carbon-
CO2 infrastructure neutral alternatives. Acquisition of the CO2 for the synthetic fuel production can be implemented in multiple
Synthetic natural gas ways. This work introduces a node-based model to assess different implementation strategies of CO2 utilization
Carbon capture
systems, taking into account temporal effects, regional variation, and economies of scale for CO2 capture.
Carbon utilization
Intermediate storage volumes, capture costs, transport quantities, and other relevant infrastructural aspects of
the CCU system can be estimated with the model. Finland is used as a case study, focusing specifically on the
national and regional scale. CO2 capture costs are significant, being nearly four times larger than the cost of
storage in the baseline scenario (354 M€, 85 M€). CO2 sources with smaller annual emissions increases capture
costs by 14% compared to baseline. This increase in cost is comparable to the cost of transporting over a quarter
of all captured CO2 to off-site processing (varying distance, 100–400 km). Seasonal storage of CO2 is found to be
beneficial for the cost-efficient production of synthetic fuels, owing to the temporal disparity between CO2
emissions and utilization, as well as the overall cost structure of the components. Five key decision categories are
proposed for a carbon utilization system: scale, type, units, location, and technological decisions. These may be
applied to describe any carbon utilization system, helping to form a more comprehensive picture of a future
energy system, where carbon is widely used as a raw material.

1. Introduction
The global share of wind and solar energy of electricity production
in 2050 is predicted to be in the range of 14–50% [1,2]. The operational
flexibility of gas turbines makes them a promising alternative to work
alongside renewable sources in a reliable future energy system [3,4]. If
the natural gas used in these turbines is substituted with sustainable
synthetic natural gas (SNG), climate effects could be decreased even
further. It is also possible to substitute fossil fuels in the transportation
sector with synthetic alternatives. Another benefit is that the produc-
tion peaks of variable renewable electricity can be balanced through
power-to-gas (P2G) processes. This is made possible by the load bal-
ancing of electrolysers, which could convert surplus renewable elec-
tricity into hydrogen. The hydrogen is then combined with CO2 to
produce hydrocarbon products. This P2G pathway offers a carbon-
neutral option for powering various parts of the energy system with
existing infrastructure, while also offering integration benefits with
intermittent renewable energy production. The technology can also be
freely scaled from the small-scale activities of individual communities
all the way to ambitious national targets that have a significant impact
on climate.
In simple terms, two main conditions need to be fulfilled by SNG

⁎
Corresponding author.
E-mail address: hannu.karjunen@lut.fi (H. Karjunen).

http://dx.doi.org/10.1016/j.apenergy.2017.07.111
Received 6 May 2017; Received in revised form 11 July 2017; Accepted 26 July 2017
Available online 02 August 2017
0306-2619/ © 2017 Elsevier Ltd. All rights reserved.
H. Karjunen et al.
Nomenclature
CAPEX Capital Expenditure
CARSON CARbon SOurce Nodal model
CCS Carbon Capture and Storage
CCU Carbon Capture and Utilization
CCUS Carbon Capture, Utilization, and Storage
CHP Combined Heat and Power
production in order to minimize climate effects: (I) the SNG manu-
facturing process uses biogenic1 or atmospheric carbon, and (II) the
electricity consumed during production does not contribute to climate
change [5]. This study is oriented more toward the first condition, in-
vestigating the various alternatives and issues connected with acquiring
CO2 for P2G processes in large quantities. P2G processes can be applied
to produce a wide range of products and fuels, but in this paper we are
focusing on the production of SNG. However, the applied modeling tool
in this work can also be used for other carbon capture and utilization
(CCU) processes, such as the production of methanol, plastics, en-
chanced oil recovery, refrigerants or even carbonated drinks.
Generally, all CCU processes require CO2 as a raw material. It is thus
beneficial that the best potential CO2 sources are identified and located,
and that a cost-effective way of implementing CCU within the existing
energy system is found. Potential CO2 sources have been investigated in
the literature. For instance Reiter and Lindorfer [7] identified the best
sources of CO2 in Austria, Middleton et al. [8] mapped CO2 sources for
the United States, and Teir et al. [9] analyzed CO2 sources in the Nordic
countries in the context of carbon capture and storage (CCS). Most of
the recent studies have focused on fossil CO2 sources and especially CO2
capture from power plants. However, many industrial processes such as
steel, ethylene, or cement production, which are difficult to make
carbon neutral, could offer higher value for the captured CO2 [10].
Previous studies have also demonstrated that models can be used to
cover various scales [11–13]. Most importantly, this approach helps in
assessing the effects of regional variation in CO2 sources.
Carbon capture, utilization, and storage (CCUS) integration typi-
cally relies on macro scale system optimization, which often uses either
economic or environment values as objective criteria. For instance, von
der Assen et al. [14] introduced an environmental-merit-order curve for
CO2 , which they used to rank the various CO2 sources of CCU in Europe.
Hasan et al. [13] presented a minimum-cost supply chain network for
CCUS, focusing on enhanced oil recovery as the available utilization
path of CO2 . Han and Lee [12] used a scalable and stochastic infra-
structure model to financially optimize carbon sequestration, utiliza-
tion, storage, and transportation with respect to a specific climate goal
for the eastern part of South Korea. [15,16]. Recently, Lee et al. [17]
introduced a CCS network-optimization tool that incorporates both
economic and environment values, while also taking into account de-
cision-making risk. As can be seen, many studies in this field are con-
cerned with geological sequestration of fossil CO2 , which is funda-
mentally different from the idea of recycling the existing carbon in the
natural carbon cycle that this work looks into. The carbon recycling
approach offers a permanent carbon neutral solution, as the CO2 that is
emitted would be later used to produce hydrocarbon products, which
could eliminate the need for introducing additional CO2 from the fossil
reserves into the atmospheric carbon cycle.
CCUS transport networks have also been studied extensively, fo-
cusing on transport method costs [18–20], transport configurations
[21,22], or more broadly on the overall chain [11,23,24]. Variable
1
Biogenic carbon is defined here as carbon “directly resulting from the combustion,
decomposition, or processing of materials other than fossil fuels, peat, and mineral
sources of carbon through combustion, digestion, fermentation, or decomposition pro-
cesses” [6].
34
Applied Energy 205 (2017) 33–43
DAC Direct Air Capture
HHV Higher Heating Value
LHV Lower Heating Value
OPEX Operating Expense
P2G Power to Gas
RES Renewable Energy Sources
SNG Synthetic Natural Gas
uncertainties have been addressed by some studies [17,25,26], while
others account for temporal effects [26–29]. The temporal scale of CCS
studies is typically several years. This time resolution is adequate for
determining when cavern storages become available or unavailable, or
when emissions sources reach the end of their lifetime. However, to
produce CCU-based fuels or products, shorter timespans should also be
considered. These CCU processes include chemical reactions, which are
inherently more sensitive to variations in input flows, and typically
require buffer storage to guarantee a stable production process.
There are many possible configurations of P2G deployment, which
refer to the various ways of placing P2G units within a country and
matching these, temporally and spatially, with potential CO2 sources.
This entails the sizing and placement of individual plants, intermediate
storage volumes, transport quantities, and the overall scale of im-
plementation. These factors are temporally linked, so it is necessary to
investigate these systems using a time resolution that can account for
daily and weekly variations in storage levels. Thus, suitable modeling
tools are required to comprehensively describe the system, and that
have the required flexibility to perform analyses for a wide range of
specifications and with different modeling precision.
This study outlines a large-scale P2G production chain for Finland,
using the developed CARSON-model (CARbon SOurce Nodal network
model). CARSON is developed as a platform for evaluating various CCU
infrastructure configurations, which can be compared using economic
and other quantitative values. The combination of an energy system,
P2G technology, and a high temporal resolution enables a more so-
phisticated investigation of the interactions within the system. The
background for this study is provided by a 100% renewable energy
scenario for Finland in 2050, resulting from the economic optimization
of an energy system using the energyPLAN model [30]. Only biogenic
CO2 sources are considered to maintain the 100% renewable vision. For
simplicity, the effects and limitations of the electric grid are left outside
the scope of this study. Furthermore, this work does not incorporate
mathematical optimization routines in a strict sense, but rather inspects
the outcome of preselected scenarios.
One important aim of this work is to evaluate the CO2 storage vo-
lumes and transport capacities in different capture and utilization sce-
narios. Fundamentally, the issue is about defining the necessities of a
CCU system, which is critical information for a successful im-
plementation of the technology. Investigating the roles and importance
of storage, capture units, and other crucial aspects of the system gives a
basis for the economic evaluation of system costs, which can be utilized
in decision-making.
2. Materials and methods
CARSON is a node-based, scalable material balance tool that can
model the flow of carbon through the energy system and derive costs
for CO2 transportation, capture, and storage. CARSON has been speci-
fically designed for evaluating the implementation strategy and in-
tegration benefits of CCU within the existing energy system. The model
is implemented using the MATLAB programming language, which was
chosen on the grounds of being rapidly adjustable and extendable to
different conditions and research needs. The CARSON model distin-
guishes itself from any other previously introduced CCU models,
H. Karjunen et al.
especially due to its temporal modeling capability. Because the model
can be closely associated to any energy system, it offers a new approach
to hourly balancing of CO2 availability and demand. The model can
easily adapt to different geographical areas, and studies can be scaled
from the level of individual plants all the way to continent-wide studies.
This study represents the first comprehensive application of the
CARSON model with a full description of the model mechanics [31].
2.1. System specifications
As previously mentioned, an energy scenario provides the necessary
background information for the CARSON model. This scenario is used
to estimate the total CO2 emissions produced by the energy sector, as
well as to derive a quantity for CO2 necessary for utilization purposes.
In the energy scenario, roughly 40% of the total produced electricity is
used for P2G. This results in a system that produces roughly 30 TWh of
SNG annually. The SNG is being used by three sectors: industrial heat
and electricity generation (53%), residental use, district heat generation
and cogeneration (40%), and the transportation sector (7%).
Different systems are assumed to produce or consume CO2 , which in
this work are categorized into industrial sites, biogas plants, P2G fa-
cilities, or combined heat and power (CHP) plants. A database of CO2
emissions is used to assess the amount of CO2 available from each ca-
tegory. Industrial sites include pulp (68% of total fossil and biogenic
emissions), steel (15%), oil refining (10%), cement, lime and chemicals
(5%), and waste incineration (2%). The CHP category is also assumed
to include facilities that only produce district heat, that is, without the
co-generation of electricity and heat.
Fig. 1 illustrates the quantities of CO2 that are captured and used for
the production of SNG in the baseline scenario of this work. Note that
only a portion of the CO2 is captured and directed to SNG production.
All of the CO2 from biogas is used for SNG production, and equal
amounts of CHP and industrial CO2 are used (4 Mt each). A relatively
larger portion of the available emissions originates from industrial
sources. Naturally, this distribution represents just one potential solu-
tion, which was assumed for this work. The higher fraction of industrial
emissions is based on a view of industrial sites having better suited
annual operational hours, scale of units, and potential integration
benefits with P2G, such as oxygen demand.
Industrial facilities and CHP plants are assumed to be run with two
fuel types: biomass and SNG. CHP gets 60% of the fuel energy from
biomass, and industry 68%. Part of the carbon contained in the fuel
mixture is recycled to produce the SNG. The carbon can originate fully
from biomass, or some of it can be recycled from the used SNG. Fig. 2
illustrates a potential pathway for CHP, where the carbon embedded in
the SNG is being recycled, and biomass is used only to cover the carbon
losses that are assumed to occur during the capture and manufacturing
processes.
2.2. CO2 balance
Fig. 3 depicts the overall flow of the CARSON operation procedure.
The total installed capacity of a system and its distribution to calcula-
tion nodes can be defined separately. Additionally, the operation
strategy of these systems for each time step is required. By combining
this information, a CO2 balance can be calculated for each calculation
node and time step.
A positive balance means that CO2 can be stored for later use,
whereas a negative balance means that CO2 must be extracted from
existing storage. Nodes may also request to transfer CO2 from the sto-
rage of neighboring nodes. Physical location, storage size, and rules
describing the availability of different transport methods can be spe-
cified uniquely for each calculation node.
The overall cost of the system consists of storage, transport, and
capture costs. All investments costs are transformed into annual peri-
odic payments, so that the total costs of different configurations are
35
Applied Energy 205 (2017) 33–43
comparable. All the costs have been indexed to 2010.
2.3. Spatial distribution
The node-based approach enables the model to be used with dif-
ferent levels of detail. In this work, two levels are used: a national
study, where Finland is divided into eight calculation nodes, as shown
in Fig. 4, and a more detailed study of the southeast region of Finland.
Initial survey of the area reveals 84 facilities, which are reduced to ten
nodes that have an active role in the carbon utilization scheme. The ten
nodes were chosen manually by ruling out smaller heat plants and back-
up plants, which are likely to have a fewer annual full load hours.
In the regional study, each node represents a single site, such as a
power plant or an industrial site. Conversely, in the national study, a
single node represents a group of plants within a given region. This
region is then idealized as a point location, with extensive properties,
such as CO2 storage size.
The total installed capacity of each CO2 source or sink type can be
specified as thermal power (th ), electric power (el ), or directly as a mass
flow rate of CO2 . In the case where the input is electric power, it is first
converted to thermal power by dividing by the electric efficiency of the
process. Then, the nodal distribution is calculated using Eq. (1).
Cn,x = Cx ·cn,x , (1)
where C is the installed capacity of technology x in node n, and c is the
fraction of total capacity present in the investigated node.
The values for total capacity (C) originate from the background
energy scenario, and are introduced in Table 1. These values reflect the
total national capacity in Finland in 2050. The geographical distribu-
tion of CO2 sources (c) for the national model scale has been estimated
using a CO2 emission database, which consists of historical individual
site emissions. Because the emission data apply to the current energy
system, and we require a distribution for 2050, a modified emission
distribution is calculated as a weighed average of two known dis-
tributions, a and b.
an,x bn,x
cn,x = 0.75 + 0.25 .
∑n an,x ∑n bn,x (2)
The industrial source capacity in 2050 is distributed according to
current emissions, weighting biogenic emissions by 75% (a) and fossil
emissions by 25% (b). This reflects the fact that, in future, it is probable
that industrial sites will not change location, but greater biogenic
emissions will be available from sites that currently produce fossil
emissions. Similarly for CHP, the current biogenic emission distribution
is weighted by 75%, and the estimated population distribution in 2040
is weighted by 25%. This, in turn, reflects the shift of demand in district
heat to more populated regions. The year 2040 is used for the popu-
lation distribution because no reference was found for a later popula-
tion distribution in Finland. These modifications to distributions are
examples of the flexibility of the model in adapting to various condi-
tions that could be envisioned. For instance, the availability of biomass
Fig. 1. Flow of CO2 from the various sources to SNG production. Unit is millions of tons of
CO2 .
H. Karjunen et al.
Fig. 2. Flow of a hypothetical kilogram of fuel through the CHP system.
Fig. 3. General CARSON calculation procedure.
Fig. 4. Map with the eight regions used in the high-level study, and the southeastern
region of Finland from the in-depth study.
could restrict the increase of CHP distribution in populated regions, or
heat pumps and other power-to-heat solutions could be used. In this
case, these can be reflected in the model by a new choice of parameters.
P2G distribution is an input variable of the model, so the details of
this variable are explained later in Section 3, together with scenario
descriptions. The role of biogas is considered so minor that its dis-
tribution has been accepted without modification. The obtained nodal
distributions of capacity are shown in Table 2.
2.4. Temporal distribution
The temporal distribution is calculated as
Cn,t ,x = Cn,x ·sn,t ,x ,
where t is the time step and s is the operation status, which is defined as
a dimensionless number between zero and one. This describes the
portion of capacity in use during each time step, compared to the total
installed capacity. Fig. 5 shows how the operation status varies for each
Applied Energy 205 (2017) 33–43
system during the year. P2G operation fluctuates heavily on a daily
basis, which causes the erratic behavior observed in the figure. This can
be connected back to the daily fluctuation in solar and wind energy
production. Each node is assumed to have identical operation curves
throughout the year, so there is no geographical variance in operation.
2.4.1. CO2 balance
The quantity of CO2 (Q) produced or consumed during each time
step is calculated as shown in Eq. (4). This step is unnecessary if the
input capacity is given directly as a mass flow rate of CO2 , which is
emphasized in the equation by using the term Cth to refer to thermal
capacity, instead of the general capacity C.
2
⎛ Cth,n,t ,x ⎞
Qn,t ,x = ∑ ⎜pf ,x · q ·Yf ·Δt ⎟.
f =1 ⎝ f ⎠ (4)
The term p refers to the proportion of a particular fuel f, which in
this work is either biomass or SNG. The term q is the lower heating
value of the fuel, and Y corresponds to the yield of CO2 from one
kilogram of fuel (i.e., emission factor). Finally, Δt is the time step size,
which in this work has been fixed to be one hour. Values for p,Y , and q
are given in Table 1.
To obtain the final available quantity of CO2 , the obtained quan-
tities need to be multiplied by the efficiency of the capture system. A
value of 85% is used for all capture devices, regardless of the CO2
source, and a value of 1/75% is used for P2G technology to account for
the CO2 losses in the SNG manufacturing chain. For each node n and
time step t, the CO2 balance can be calculated by taking the sum of all
the CO2 sources and sinks x present in the current node.
2.5. Transport, storage and CO2 capture
Fig. 6 depicts a simplified version of the algorithm used by the
model. Each node has a CO2 storage level, which is updated based on
the calculated CO2 balance during each time step. Next, the model can
either directly advance to the next time step, or alternatively, it can
attempt to transfer CO2 from another node. This happens in the case
where the node’s own storage reaches its minimum level. Exporting
nodes are searched in the order of shortest distance, and a successful
transfer requires that the exporting node has an adequate storage level,
and that transports are allowed between the nodes. If the checks fail,
another node is tried, or ultimately, the production of P2G is decreased
if there is no CO2 available for the process. This procedure is performed
for every time step and calculation node.
CARSON calculates the transportation cost on an annual basis be-
tween each node pair with every available transport method, provided
that the particular node pair can transfer resources between each other.
Then, the transport method with the minimum total cost is chosen for
each node pair. Pipe, truck, and rail transport are the possible options
used in this work. These transport methods are explained briefly in the
following sections.
2.5.1. Pipe transport
Pipe costs are calculated as a function of pipe length L, in kilo-
meters, and pipe diameter D, in nominal pipe size (NPS). The capital
expenditures (CAPEX) for the pipe are calculated as shown in Eq. (5),
which is derived from regression analyses of published natural gas pi-
peline project costs from the United States, as explained by Mccoy and
Rubin [38]. Relevant regression coefficients are listed in Table 3.
log(CAPEX) = a0 + a6logL + a7logD. (5)
(3)
The pipe diameter is calculated iteratively, based on the annual
quantity of CO2 transferred and a fixed pressure drop. Relevant para-
meters are given in Table 4. The haversine formula is used to calculate
the pipe length, based on the latitude and longitude coordinates of the
calculation nodes. The calculated costs are converted to euros, and the
36
H. Karjunen et al. Applied Energy 205 (2017) 33–43

Table 1
Model variables and data sources.

Data Type Value Unit Comment Reference

Installed capacity (C) Industry 1.46 GWel [30]

CHP 7.05 GWel [30]
Biogas 63.5 t CO2/h [32]
P2G 8.4 GWth,SNG Thermal content of produced SNG [30]

Operation status (s) Industry [0…1] – Value corresponds to the proportion of nodal capacity in operation, defined separately [30]
District heat [0…1] – for each time step. All nodes are assumed to have identical operation. [30]
Biogas [0…1] – Synthetic data
P2G [0…1] – [30]

Emission factor (Y) Biomass 1.8 kgCO2/kgBio Dry [33,34]

SNG 2.7 kgCO2/kg SNG [33,34]

Heating value (q) Biomass 18 MJ/kg Lower Heating Value (LHV), dry [33,34]
SNG 49 MJ/kg LHV [33,34]
Hydrogen 142 MJ/kg Higher Heating Value (HHV) [33,34]

Share of energy from Industry 68 % Remainder is SNG [30]

biomass (p)
CHP 60 % Remainder is SNG [30]

Efficiency Carbon capture 85 % kg captured/kg processed [9]

Carbon conversion 75 % kgCO2,100%conversion/kgCO2 [35]
Electricity 40 % Jel /Jth [36]
generation
P2G 52 % Jth,SNG,LHV / Jel [37]

Transport costs Train See Table 5 [12]

Pipe See Tables 4 and 3 [38]
Truck See Table 5 [12]

Other costs Carbon capture See Table 7 [25]

Storage See Table 6 [39,24]
P2G See Table 7 [40]
a
Distribution of capacity Industry % See Table 2 Database
(c)
CHP % See Table 2 [41], Databasea
Biomass % See Table 2 Databasea
P2G % Model input parameter

a
The referred database contains reported emission statistics of existing plants, and was originally compiled from the European Pollutant Release and Transfer Register (E-PRTR) 2007
data. The database has been updated in 2010 [9], and again in 2012 to cover the E-PRTR 2010 statistics [24]. In 2015–2016, the data has been updated by the authors of this paper, using
data from The Finnish Energy Authority [42], as well as [32,43] and other manual additions from companies’ yearly reports or similar sources.

Table 2 load of a single vehicle, Q load , and taking the ceiling of this value.
The predicted share of Finland’s national installed capacity of the major biogenic CO2 Multiplying this by the distance between nodes, L, and noting that each
sources in different regions in 2050.
vehicle needs to return after each trip, the total distance is obtained.
Node CHP Industry Biogas Unit
Q
Ltot = 2L ⎡ tot ⎤.
1 6.5 6.2 8.6 % ⎢
⎢ Q load ⎥
⎥ (6)
2 10.6 13.0 8.8 %
3 12.2 8.7 14.9 % Next, the total time taken to perform all transports, ttot , is affected by
4 12.6 7.8 14.5 % vehicle speed, v, and loading and unloading time of the vehicle, tload .
5 10.3 34.3 13.4 %
6 26.7 7.5 22.9 % Ltot Q
7 14.7 9.3 13.3 %
ttot = + ⎡ tot ⎤ tload.
v ⎢
⎢ Q load ⎥
⎥ (7)
8 6.2 13.1 3.6 %
Finally, the number of vehicles is obtained by dividing the total time
Total 100 100 100 %
by the availability hours of a single vehicle, tava . Investment cost is
calculated by multiplying the number of vehicles N by the investment
cost index for pipes, valves, and fittings is used to calculate the costs for cost of a single vehicle, and the costs are transformed into annual
2010 [44,45]. payments. The costs are expressed for 2010 [12]. Relevant parameters
used in the calculations are shown in Table 5.
ttot
2.5.2. Vehicle transport N= .
tava (8)
Transport costs for truck and rail transport are calculated using the
method described by Han and Lee [12], which considers the costs of
investment, fuel, labor, maintenance, and general expenses. To calcu- 2.6. Storage and capture costs
late the costs, the number of required vehicles needs to be known.
Consequently, to calculate the number of vehicles, the total distance The CARSON model calculates capture and storage costs using data
traveled, Ltot , is required. Number of shipments can be calculated by from a reference plant, which is derived from existing literature. A
dividing the total quantity of goods transported, Qtot , by the maximum 3000 m3 cylindrical tank is used as a reference for the storage costs

37
H. Karjunen et al. Applied Energy 205 (2017) 33–43

P2G
1

Operation status 0.8

0.6

0.4

0.2

0
CO 2 Sources
1

0.8 Industry
Operation status

0.6

0.4 Biogas

0.2 CHP

0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan
Fig. 5. Operation status s of each considered system during the year.

Table 5
Transport costs for vehicles.

Variable Truck Train Unit

Availability (tava ) 6570 4380 h/a

Investment 374,321 487,577 €/unit
Fuel cost 0.57 1.34 €/km
General expense 3788 3157 €/a
Hourly cost 27.55 27.55 €/h
Loading time (tload ) 2 12 h/trip
Maintenance 0.133 0.0886 €/km
Capacity (Q load ) 22 80 t CO2/trip
Speed (v) 55 45 km/h
Interest rate 10 10 %
Lifetime 10 10 a

Fig. 6. Simplified procedure for transport and storage algorithm.

increase in cost of storage tanks, compressors, heat exchangers and
Table 3
other equipment corresponds roughly to the increase in capacity raised
Coefficients used to calculate pipe transport costs. to the power of six-tenths. The value of the exponent can be adjusted for
different industries and equipment, but for this work, the value of 6/10
a0 a6 a7 was considered to give a reasonable approximation [12,46].
Material 3.112 0.901 1.59
The model calculation process is divided into two parts. First, we
Labor 4.487 0.82 0.94 calculate the number of units N present in a node n for source type x.
Right-of-way 3.95 1.049 0.403 This step is not done for the small-scale model, because in that case,
Miscellaneous 4.39 0.783 0.791 each node represents a single plant and Nx ,n is always equal to one.
Otherwise, the total annual quantity of CO2 in node n obtained from
source type x is divided by the average annual emission quantity of the
Table 4
corresponding source type, Qave . Furthermore, a minimum of one unit
Additional variables used to calculate pipe transport costs.
needs to exist. The average emission quantity, Qave , represents the ty-
Pipe lifetime 30 a pical size of a CO2 source, and varies between scenarios, as discussed
Interest rate 15 % later in Section 3.
CO2 density 900 kg/m3
Viscosity 110 μPa·s Nx ,n = max(1,Qx ,n/ Qave,x ). (9)
Pipe roughness 0.0457 mm
Outlet pressure 10.3 MPa
Once the number of units present in node n has been determined,
Inlet pressure 13.8 MPa the CAPEX cost is calculated using the six-tenths rule, as shown in Eq.
Operational expense 3250 US"$"2004 /km·a (10). The term Q is the quantity used to describe the factor that de-
termines the cost of a unit, for example, tons of CO2 that can be stored
or tons of CO2 captured. The subindex “ref” is used to distinguish the
[39]. Cavern-based underground storages have lower costs in larger reference unit, the size and cost of which we know beforehand.
scale [24], but built tanks have more freedom in terms of placement.
0.6
Thus, undeground storages were not considered in this study. Capture Qx ,n ⎞
CAPEXn,x = Nx ,n ⎜⎛ ⎟ CAPEXref,x.
costs are based on an amine-scrubbing process for all of the considered Q
⎝ ref,x ⎠ (10)
CO2 sources [25]. Furthermore, the model uses the six-tenths factor rule
to estimate the effect of unit scale benefits. The rule states that the Operating expenses (OPEX) are simply calculated either as a per-
centage of CAPEX, or by multiplying of a constant by the cost factor, Q.

38
H. Karjunen et al. Applied Energy 205 (2017) 33–43

The values used in these calculations are shown in Tables 6 and 7. The comparison is not worthwhile.
storage costs are indexed from 2008 to 2010 according to the cost index
of heat exchangers and tanks [44,47]. 4.1. Scenario costs

2.7. Levelized cost of production Fig. 10 shows the total cost of all of the scenarios. In national scale,
a low system cost is achieved when CO2 is captured mostly from in-
Levelized cost of SNG production, LCOSNG , is used for a fair dustry, or when larger unit sizes are used. In both cases, the difference
comparison of the results. [48] In this work, it is used to compare can be explained by the decreased capture cost.
systems producing different quantities of SNG. The cost figure is ana- The lower cost of the IND60 is explained by two factors. First, in-
logous to levelized cost of electricity, and is calculated as: dustrial units have a larger capacity factor. Industrial sites are assumed
CAPEX a + OPEX to have about 6411 full load hours during a year, whereas CHP only had
LCOSNG = .
Eth,SNG,a (11) 1990. Second, industrial units are assumed to have a lower investment
cost for capture devices (Table 7). The other scenario that reaches a low
CAPEXa is the CAPEX cost of the system, which has been converted total cost is large, where the lower capture cost originates from the six-
into annual payments. The term ESNG,a is the annual thermal energy of tenths rule. The opposite effect can be seen with smaller unit sizes.
the produced SNG. The unit of the levelized cost is €/MWh. Interestingly, the west scenario has approximately the same total
cost as the small scenario. The average cost of capture is 43 €/ton in the
3. Scenarios west scenario, which increased to only 49 €/ton in the small scenario.
When the total quantity of CO2 captured is measured in millions of tons,
The main goal set for the system is that 30 TWh of SNG must be even relatively small differences in capture prices are significant on a
produced annually. According to the energy scenario, this amount system scale. These results suggest that the costs of transportation could
would be enough to completely substitute natural gas in various ap- potentially be covered by the savings achieved from using cheaper CO2
plications that would continue using gas in 2050. To achieve this, the sources.
total captured CO2 amount has been set slightly larger than demand in
all of the scenarios. Similarly, the size of storage is determined so that
4.2. CO2 balance and storage
there is always CO2 available for the P2G process. These choices are
made to guarantee that the desired production goal is met.
The largest imbalance between CO2 capture and consumption is
Altogether, 12 scenarios are used to illustrate the various possible
observed in the autumn time, as seen in Fig. 11. The reason for this is
configurations. Half of these are used in the national scale model, and
the assumed operation characteristics of the different systems, which
the other half in regional scale. As an output of a scenario, the model
can be traced back to the energy scenario. The amount of energy pro-
gives financial values corresponding to the total system cost. The de-
duced from wind power reaches its peak in autumn, which is reflected
scriptions of these scenarios are shown in Table 8, and Fig. 7 shows
as a surge in P2G production. Another, slightly less drastic spike is also
graphically how the nodes are distributed in the regional scenarios.
observed in the spring. However, all nodes are assumed to have iden-
The base scenario distributes the P2G capacity based on available
tical operation curves, which does not take into account regional var-
CO2 emissions, which also happens to favor regions that currently have
iations in wind power production, or other related fluctuations.
natural gas network coverage. This would undoubtedly ease the dis-
The storage levels of all calculation nodes in the base scenario are
tribution of the final product, SNG. In contrast, the west scenario places
shown in Fig. 12. A minimum stored amount of 5 kt is set. The total
the P2G capacity on the western coast, which is also the part of Finland
storage size is observed to be 1.3 Mt, whereas the total captured CO2
with the highest wind power potential. This scenario is chosen for two
quantity is 8.2 Mt. Storage size is affected by the capacity factor of the
reasons. First, P2G plants could require or benefit from a connection to
CO2 sources, as can be expected. Another factor that can affect to the
a wind turbine, which could restrict the placement of P2G units in this
storage size is the location of the P2G units. The west scenario storages
region. Second, we want to determine whether it is feasible to transport
were about 20% larger than in the base scenario, which is caused by the
CO2 over long distances across regions and on a large scale. This bulk
mismatch between the location of CO2 sources and P2G facilities.
transport situation is reached because the same CO2 sources are used,
Decreasing the storage capacity from the current level eventually
and the location of these sources no longer matches the P2G production
leads to reduced SNG production, as CO2 is not available for the pro-
locations.
cess. For the base scenario, the levelized cost of SNG production is
Two scenarios, large and small, are used to investigate what happens
111.5 €/MWh, which line with other studies [48]. When the CO2 sto-
when the average unit size of CO2 sources change. These scenarios are
rage capacity is decreased by 1 Mt, the levelized cost of production
used to describe a strategical difference in carbon acquisition, either as
changes to 124.6 €/MWh. Even though the absolute costs of storages
a centralized system with large unit sizes, or as a distributed system
decreases by 62 M€ on an annual level, the CAPEX cost of P2G and CO2
with smaller units. Fig. 9 illustrates the assumed sizes of these units,
capture remain unchanged. Thus, when the annual amount of SNG
given in annual quantities of CO2 produced. The assumed average unit
produced decreases by 3.67 TWh, due to the smaller CO2 storage, the
size used in the base scenario is shown with a solid line and marked
levelized cost of the produced SNG increases (see Eq. (11)). Under these
with “Default unit size.” and the lower and upper bounds of the unit
conditions, having a larger CO2 storage is more economical. One way of
sizes are shown with a dashed line (small and large scenarios). These are
maintaining the same SNG production level with lower storage capacity
used in the model to derive the cost of CO2 capture, by assuming that all
is to increase CO2 capture. However, this also results in an larger
units are of the same size (Q in Eq. (10)).
Table 6
4. Results Parameters for storage cost calculations [24,39].

Table 9 presents the components of the total system costs for the Reference size 3000 m3
base scenario. The estimated cost of P2G is clearly higher than that of CO2 density 1100 kg/m3
storage, capture, or transport. Capture costs are still quite significant, CAPEX 2.94 M€
OPEX 1 % of CAPEX
but the impacts of transfers and storage are quite minor. Similar results
Lifetime 30 a
were obtained in the work of Han and Lee [12] and Hasan et al. [13], Rate 7 %
but the modeling environments are so different that a detailed

39
H. Karjunen et al. Applied Energy 205 (2017) 33–43

Table 7 which results in an absolute lower quantity of transfers required. The

Parameters for calculating system capture costs [25]. same principles can be used to explain the lower transport costs when
there are fewer P2G facilities.
Industry CHP Biogas P2G
On a national scale, only transfers across region boundaries are
CAPEX 24.3 123.4 13.3 M€/unit studied. Each region can also have internal resource transfers, as we
CAPEX 1400 €/ kWel showed for one particular regional area. The quantity of transfers is
OPEX 0.94 1 4.6 € snt/ kgCO2 expected to vary between regions, depending on factors such as the
OPEX 5 % of CAPEX availability of suitable rail connections, the number and size of CO2
Lifetime 30 30 30 20 a
Rate 14.8 14.8 14.8 5 %
sources, and the physical placement of units.
Reference size 100 400 100 n/a kt CO2/a

levelized cost than in the base scenario. In our approach, we assumed
that the P2G production is not flexible, but rather it has to be operated
at designated times, or not at all. The reasoning for this is the avail-
ability of hydrogen, which, in turn, is driven by the purchase price and
availability of renewable intermittent electricity. The storage of hy-
drogen is considered to be more complex than the storage of CO2 , which
could limit the available hydrogen buffer for operational flexibility.
4.3. Transport
As can be seen from Table 10, pipe transport is the dominant form of
transfer in the regional scenarios. The larger the annual quantity of
transfers is, the more economical pipe transport becomes. Thus, sce-
narios with a centralized structure, such as rlarge1P2G, are able to have
lower total costs. Smaller CO2 sources caused the transfers to shift to
truck and rail transport, because these methods are preferable if the
annual transfer quantities remain low. In the west scenario, train is the
dominant form of transfer. This is caused by the longer transport dis-
tance, because the annual quantity of transfer would need to be higher
for pipe transfer to be economical. Using optimally constructed trunk
lines could increase the competitiveness of pipe transport [22].
Transport costs decrease with larger CO2 capture units. One reason
for this is that with larger unit sizes, the transfers are concentrated
among a smaller number of nodes. This enables economical pipe and
rail transport methods, because the transferred quantities are higher on
a node-to-node basis. Another reason has to do with P2G placement.
P2G units are assumed to be placed beside the largest CO2 sources,
5. Discussion
Previously, Middleton and Bielicki [11] identified seven key deci-
sions that should be considered simultaneously by a comprehensive
CCS infrastructure model: “(1) how much CO2 to capture (2) at which
sources; (3) where to construct pipelines and (4) of what size; (5) which
reservoirs should store CO2 and (6) how much to inject; and (7) how to
distribute CO2 from the dispersed sources through the network to the
reservoirs.” However, these decisions are not adequate to cover a CCU
system comprehensively, owing to the larger number of components
and options available.
Here, an alternative classification is given, where the CCU system is
divided into four system components, namely supply, demand, storage,
and transport. For each of these components, there are five decision
categories: scale, type, units, location, and technology. As an example, the
scale category includes the total quantity of CO2 captured for the supply
component of the system. A matrix of the decisions and components is
given in Table 11, including examples.
The presented matrix form, in conjunction with the data obtained in
this work, shows that the decisions are interconnected, both within
each component and across components. For example, power plants
and industrial sites generally produce larger quantities of CO2 than
biogas sources, which links the type and units categories. Furthermore,
the operation mode of these CO2 sources affects the size of the required
intermediate storage, which becomes apparent when baseload units are
compared to peaking plants, for instance. This represents the inter-
connectivity between supply and storage, which in this case also happens
to be time-dependent.
Clearly, to make any decisions considering transport and temporary

Table 8
Scenario descriptions.

Name Origin Description Unit size kt CO2 / a Number of P2G

units
Industry CHP

National scenarios
Base Roughly equal amount of CO2 captured from CHP and industry, and minor quantities from biogas. No CO2 800 548 8
transport, because P2G capacity is distributed according to captured emissions
West Base P2G capacity is shifted to western coast of Finland, which means that CO2 needs to be transported across 800 548 8
regions. Same CO2 sources are used as in base scenario. See Fig. 8
IND60 Base The share of CO2 captured from industry is increased to 60% 800 548 8
CHP60 Base The share of CO2 captured from CHP is increased to 60% 800 548 8
Large Base Units used to capture CO2 are assumed to be larger. See Fig. 9 1600 1100 8
Small Base Units used to capture CO2 are assumed to be smaller. See Fig. 9 400 270 8

Regional scenariosa
rBase (Base) CO2 is captured from three industrial and two CHP plants. Each plant has its own P2G plant beside the CO2 504 356 5
source. Unit sizes are based on current historical plant emissions. The quantity of P2G production is based on
national base scenario
rBase1P2G rBase A single P2G unit is used, and CO2 is transported from nearby sites to satisfy its CO2 demand 504 356 1
rLarge rBase CO2 is captured from two large plants: one industrial and one CHP plant. The plants are based on real 1522 716 5
facilities, but the emissions have been scaled up
rLarge1P2G rLarge A single P2G unit is used. The unit is placed in the same location as the industrial site. A single CHP unit 1522 716 1
captures and transports additional CO2 to satisfy the demand
rSmall rBase Unit sizes are smaller. Five industrial units and four CHP units are used to capture CO2 305 179 5
rSmall1P2G rSmall One P2G unit is used. It is placed beside the largest available industrial site. Transport is required from other 504 356 1
CO2 sources

a
The unit sizes for regional scenarios are calculated averages. Each site actually has a unique unit size.

40
H. Karjunen et al. Applied Energy 205 (2017) 33–43

rbase rbase1P2G rlarge rlarge1P2G rsmall rsmall1P2G

Biogas
CHP
Industry

P2G

Fig. 7. Calculation nodes of the regional scenarios. Top row shows the location and relative size of CO2 sources, and bottom row shows the same information for P2G facilities, which act
as CO2 sinks. Each column represents a scenario.

National scale Regional scale

500 120

400 100

80
300
60
200
40
100 20

0 west 0
small
CHP60
base
large
IND60

rsmall1P2G
rsmall
rbase1P2G
rbase
rlarge
rlarge1P2G
Storage
Capture
Transport
(a) west (b) base
Fig. 10. Total system cost of the scenarios.
Fig. 8. The distribution of P2G capacity in the west and base scenarios. The shown values
represent the fraction of total P2G capacity in each investigated region.
storage, temporal information should be available. Thus, modeling
CHP Industry tools should also include a temporal aspect, especially when the mod-
80 30 eling accuracy increases, and details about individual sites are included.
Default unit size
Default unit size

25 On the other hand, more general macro-scale models have a clear role
Number of units

60 in devising rougher drafts of the system specifications with little input

20
data.
40 15 To make any kind of scenarios or predictions of future CCUS sys-
tems, some of these decisions presented in Table 11 need to be fixed.
10
20 Based on the cost figures presented in this paper, P2G production, or
5 demand, is the component associated with the highest cost. This is re-
flected in this work by stating a fixed production goal for SNG. In
0 0
0 500 1000 1500 2000 0 1000 2000 3000 4000 reality, this choice would most likely be driven by climate goals, which
Unit size (kt CO /a) Unit size (kt CO /a) would mean that funds and relevant policies could be directed to enable
2 2
this approach, offsetting the relatively high costs of P2G production.
Fig. 9. Histogram of current CHP and industrial facility unit sizes in Finland, with the
The second highest cost component is associated with supply, which is
used average unit sizes in various scenarios.
reflected here by fixing the quantity of CO2 captured to be as low as
possible. An ultimate configuration of the CCU system is difficult to
Table 9 determine owing to the many uncertainties, but identifying the most
System costs for base scenario. critical decisions can direct our focus toward potential pathways. In-
System costs Total (M€) CAPEX (M€) OPEX (M€) €/ton
dividual investigations can then be used to observe effects in detail.
The effects of having different capture technologies was not con-
Storage 84.9 75.5 9.4 10.3a sidered in this work. There is a wide range of emerging CO2 capture
Capture 354.2 263.5 90.7 43.1b technologies (see e.g. [49]), which could potentially deliver CO2 at
P2Gc 2959.7 1823.5 1136.2
lower energy penalties and costs than the amine absorption that was
a
Calculated by dividing the total annualized cost of storage by the total quantity of considered in this work. Selected average capture costs for a single
CO2 captured during a year. technology were seen sufficient for the present study, since there is a
b
Calculated by dividing the total annualized cost of capture by the total quantity of large deviation in the reported capture costs mainly due to lack of large-
CO2 captured during a year. scale demonstration plants. It can be hypothesized that the influence on
c
The presented numbers are only a crude approximation [40].
capture cost would be even more prominent with different

41
H. Karjunen et al. Applied Energy 205 (2017) 33–43

600
Industry
500
CHP
400 Biogas
CO 2 (kt) 300
P2G

200

100

0
4 8 12 16 20 24 28 32 36 40 44 48 52
Week number
Fig. 11. Captured CO2 from different sources in the base scenario and corresponding P2G demand. Each column represents a two-week period.

350 Fig. 12. CO2 storage in the eight calculation

nodes during the year with the base scenario.
300
Stored CO 2 (kt)

250
200
150
100
50
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan

Table 10 6. Conclusion
Transportation costs and methods.
This work introduced a new modeling tool, called CARSON, which
Scenario Quantity Mt CO2 Average cost Truck % Rail % Pipe %
€/t CO2 was used to compare various CCU infrastructure configurations in
Finland, as a case study. The model includes costs for CO2 capture,
west 2.13 14 0.01 78 22 transport, and storage, while also accounting for economies of scale.
rbase1P2G 1.41 4.4 0 17 83 The unique feature of this model is the temporal production and con-
rlarge 1.12 5.2 0 37 63
sumption of CO2 , enabling the estimation of intermediate storage vo-
rlarge1P2G 0.51 5.9 0 0 100
rsmall 0.68 7.8 19 35 46 lumes and transport quantities. Additionally, a link between capture
rsmall1P2G 1.52 6.4 0 33 67 units and CCU facilities is formed, which together forms the basis for
the economic evaluation of systems, while also providing valuable in-
sights into decision-making processes.
technologies, than when considering only different unit sizes, as we did Based on the results presented in this paper, large CO2 capture units
here. In particular, the price difference between direct air capture and with relatively stable production during the year proved to be an eco-
conventional capture technologies could be rather high, which could nomical solution for CCU. The cost of capture plays a large role in total
mean that transporting raw CO2 to a production site could be an eco- system costs, and the efficient utilization of storage and transport can
nomical solution. However, this also implies that the placement of P2G have a beneficial impact on total system costs. The required capacity of
production sites would be restricted, for instance, by the proximity of a CO2 storage is dictated by the imbalance between supply and demand,
wind turbine, an existing transport infrastructure, or local consumption which can be affected by the choice of CO2 sources.
of the final product. At this point, it is still unclear how such restrictions Five key decision categories were proposed for a general CCU
would be applied in reality. Much of this uncertainty originates from system, which can be used to identify high-level implementation stra-
policy, which is still developing for carbon footprint handling and tegies. These strategies can be combined with a modeling tool, such as
electricity taxation in the context of CCU. CARSON, to perform regional case evaluations under various condi-
tions. CCU systems are associated with many uncertainties, which

Table 11
Key decisions for a CCU system.

Supply Demand Storage Transport

Scale What is the total quantity of CO2 What is the total quantity of CO2 What is the total storage volume? What is the quantity transferred?
captured? (t/a) utilized?
Type What type of CO2 source is used? What type of utilization? (SNG, What type of storages are used? Which transport method is used? (pipe,
(biogas, industry, atmosphere) polymers) (caverns, tanks) tanker)
Units What is the capacity of a single unit? (t What is the capacity of a single What is the volume of a single What is the capacity between points (number
CO2 /a) unit? storage? of tanks, pipe diameter)
Location Where are these units located? Where are these units located? Where are these units located? Between which points do these transports
occur?
Technology What capture technology is used? Which production process is used? Storage conditions? (temperature, Transport conditions? (temperature,
(sorption, membranes) (biological, catalytic) pressure) pressure)

42
H. Karjunen et al.
makes it necessary for the modeling tools to be highly adaptable and
flexible. After an initial investigation, more comprehensive decisions
could be considered, for instance, by including the consumption and
delivery of the final product, or by including additional technological
options for the system.
Acknowledgements
This work was supported by the public financing of Tekes, the
Finnish Funding Agency for Innovation, for the “Neo-Carbon Energy”
project under the number 40101/14.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.apenergy.2017.07.111.
References
[1] Greenpeace International, Global Wind Energy Council, SolarPowerEurope, Energy
[R]Evolution; 2015.
[2] OECD, World Energy Outlook 2015, OECD Publishing, oCLC: 930994298; 2015.
[3] Brouwer AS, van den Broek M, Zappa W, Turkenburg WC, Faaij A. Least-cost op-
tions for integrating intermittent renewables in low-carbon power systems. Appl
Energy 2016;161:48–74. http://dx.doi.org/10.1016/j.apenergy.2015.09.090. ISSN:
03062619.
[4] IPCC, Edenhofer O, Pichs Madruga R, Sokona Y. United nations environment pro-
gramme, world meteorological organization, intergovernmental panel on climate
change. In: Potsdam-Institut für Klimafolgenforschung, editor. Renewable energy
sources and climate change mitigation: special report of the intergovernmental
panel on climate change. Cambridge University Press, New York; 2012.
[5] Reiter G, Lindorfer J. Global warming potential of hydrogen and methane pro-
duction from renewable electricity via power-to-gas technology. Int J Life Cycle
Assess 2015;20(4):477–89. http://dx.doi.org/10.1007/s11367-015-0848-0. ISSN:
0948-3349, 1614-7502.
[6] U.S. Environmental Protection Agency. Accounting framework for biogenic CO2
emissions from stationary sources; 2011.
[7] Reiter G, Lindorfer J. Evaluating CO2 sources for power-to-gas applications – a case
study for Austria. J CO2 Utilization 2015;10:40–9. http://dx.doi.org/10.1016/j.
jcou.2015.03.003. ISSN: 22129820.
[8] Middleton RS, Clarens AF, Liu X, Bielicki JM, Levine JS. CO2 deserts: implications of
existing CO2 supply limitations for carbon management. Environ Sci Technol
2014;48(19):11713–20. http://dx.doi.org/10.1021/es5022685. ISSN: 0013-936X,
1520-5851.
[9] Teir S, Hetland J, Lindeberg E, Torvanger A, Buhr K, Koljonen T, et al. Potential for
carbon capture and storage (CCS) in the Nordic region, no. 2556 in VTT research
notes, VTT Technical Research Centre of Finland, Espoo; 2010.
[10] Middleton RS, Levine JS, Bielicki JM, Viswanathan HS, Carey JW, Stauffer PH.
Jumpstarting commercial-scale CO2 capture and storage with ethylene production
and enhanced oil recovery in the US Gulf. Greenhouse Gases: Sci Technol
2015;5(3):241–53. http://dx.doi.org/10.1002/ghg.1490. ISSN: 21523878.
[11] Middleton RS, Bielicki JM. A scalable infrastructure model for carbon capture and
storage: SimCCS. Energy Policy 2009;37(3):1052–60. http://dx.doi.org/10.1016/j.
enpol.2008.09.049. ISSN: 03014215.
[12] Han J-H, Lee I-B. Development of a scalable and comprehensive infrastructure
model for carbon dioxide utilization and disposal. Indust Eng Chem Res
2011;50(10):6297–315. http://dx.doi.org/10.1021/ie200344t. ISSN: 0888-5885,
1520-5045.
[13] Hasan MMF, Boukouvala F, First EL, Floudas CA. Nationwide, regional, and state-
wide CO2 capture, utilization, and sequestration supply chain network optimiza-
tion. Indust Eng Chem Res 2014;53(18):7489–506. http://dx.doi.org/10.1021/
ie402931c. ISSN: 0888-5885, 1520-5045.
[14] von der Assen N, Muller LJ, Steingrube A, Voll P, Bardow A. Selecting CO2 sources
for CO2 utilization by environmental-merit-order curves. Environ Sci Technol
2016;50(3):1093–101. http://dx.doi.org/10.1021/acs.est.5b03474. ISSN: 0013-
936X, 1520-5851.
[15] Han J-H, Lee I-B. Stochastic optimization model for carbon capture and storage
infrastructure under uncertainty in CO2 emissions, product prices, and operating
costs. Indust Eng Chem Res 2012;51(35):11445–57. http://dx.doi.org/10.1021/
ie3004754. ISSN: 0888-5885, 1520-5045.
[16] Han J-H, Lee I-B. A comprehensive infrastructure assessment model for carbon
capture and storage responding to climate change under uncertainty. Indust Eng
Chem Res 2013:1520–5045. http://dx.doi.org/10.1021/ie301451e.
130304091142003, ISSN: 0888-5885.
[17] Lee S-Y, Lee J-U, Lee I-B, Han J. Design under uncertainty of carbon capture and
storage infrastructure considering cost, environmental impact, and preference on
risk. Appl Energy 2017;189:725–38. http://dx.doi.org/10.1016/j.apenergy.2016.
12.066. ISSN: 03062619.
[18] Knoope M, Ramírez A, Faaij A. A state-of-the-art review of techno-economic models
predicting the costs of CO2 pipeline transport. Int J Greenhouse Gas Control
43
Applied Energy 205 (2017) 33–43
2013;16:241–70. http://dx.doi.org/10.1016/j.ijggc.2013.01.005. ISSN: 17505836.
[19] Kjärstad J, Skagestad R, Eldrup NH, Johnsson F. Ship transport—a low cost and low
risk CO2 transport option in the Nordic countries. Int J Greenhouse Gas Control
2016;54:168–84. http://dx.doi.org/10.1016/j.ijggc.2016.08.024. ISSN: 17505836.
[20] Geske J, Berghout N, van den Broek M. Cost-effective balance between CO2 vessel
and pipeline transport. Part I – Impact of optimally sized vessels and fleets. Int J
Greenhouse Gas Control 2015;36:175–88. http://dx.doi.org/10.1016/j.ijggc.2015.
01.026. ISSN: 17505836.
[21] Luo X, Wang M, Oko E, Okezue C. Simulation-based techno-economic evaluation for
optimal design of CO2 transport pipeline network. Appl Energy 2014;132:610–20.
http://dx.doi.org/10.1016/j.apenergy.2014.07.063. ISSN: 03062619.
[22] Morbee J, Serpa J, Tzimas E. Optimised deployment of a European CO2 transport
network. Int J Greenhouse Gas Control 2012;7:48–61. http://dx.doi.org/10.1016/j.
ijggc.2011.11.011. ISSN: 17505836.
[23] Dai C, Cai Y, Li Y, Sun W, Wang X, Guo H. Optimal strategies for carbon capture,
utilization and storage based on an inexact Mλ -measure fuzzy chance-constrained
programming. Energy 2014;78:465–78. http://dx.doi.org/10.1016/j.energy.2014.
10.034. ISSN: 03605442.
[24] Kujanpää L, Ritola J, Nordbäck N, Teir S. Scenarios and new technologies for a
North-European CO2 transport infrastructure in 2050. Energy Proc
2014;63:2738–56. http://dx.doi.org/10.1016/j.egypro.2014.11.297. ISSN:
18766102.
[25] Knoope M, Ramírez A, Faaij A. The influence of uncertainty in the development of a
CO2 infrastructure network. Appl Energy 2015;158:332–47. http://dx.doi.org/10.
1016/j.apenergy.2015.08.024. ISSN: 03062619.
[26] Zhang X, Duncan IJ, Huang G, Li G. Identification of management strategies for CO2
capture and sequestration under uncertainty through inexact modeling. Appl
Energy 2014;113:310–7. http://dx.doi.org/10.1016/j.apenergy.2013.07.055. ISSN:
03062619.
[27] Lee J-Y, Tan RR, Chen C-L. A unified model for the deployment of carbon capture
and storage. Appl Energy 2014;121:140–8. http://dx.doi.org/10.1016/j.apenergy.
2014.01.080. ISSN: 03062619.
[28] Tan RR, Aviso KB, Bandyopadhyay S, Ng DKS. Continuous-time optimization model
for source–sink matching in carbon capture and storage systems. Indust Eng Chem
Res 2012;51(30):10015–20. http://dx.doi.org/10.1021/ie202821r. ISSN: 0888-
5885, 1520-5045.
[29] Tapia JFD, Lee J-Y, Ooi RE, Foo DC, Tan RR. Optimal CO2 allocation and sche-
duling in enhanced oil recovery (EOR) Operations. Appl Energy 2016;184:337–45.
http://dx.doi.org/10.1016/j.apenergy.2016.09.093. ISSN: 03062619.
[30] Child M, Breyer C. Vision and initial feasibility analysis of a recarbonised Finnish
energy system for 2050. Renew Sustain Energy Rev 2016;66:517–36. http://dx.doi.
org/10.1016/j.rser.2016.07.001. ISSN: 13640321.
[31] Karjunen H, Inkeri E, Tynjälä T, Hyppänen T. Sustainable carbon sources for biofuel
production in renewable energy future. In: Proceedings, 10th international re-
newable energy storage conference, Düsseldorf, Germany; 2016.
[32] Tähti H, Rintala J. Potential for biomethane and biohydrogen production in Finland
(finnish: Biometaanin ja -vedyn tuotantopotentiaali Suomessa). University of
Jyväskylä, Jyväskylä, Finland; 2010.
[33] Raiko R. International flame research foundation. Suomen kansallinen osasto,
Poltto ja palaminen, Teknillistieteelliset akatemiat, Helsinki; 2002.
[34] Statistics Finland. Fuel classification 2016; 2016.
[35] Muroyama H, Tsuda Y, Asakoshi T, Masitah H, Okanishi T, Matsui T, et al. Carbon
dioxide methanation over Ni catalysts supported on various metal oxides. J Catal
2016;343:178–84. http://dx.doi.org/10.1016/j.jcat.2016.07.018. ISSN: 00219517.
[36] The European IPPC Bureau. Reference document on best available techniques for
large combustion plants; 2006.
[37] Minutillo M, Perna A. Renewable energy storage system via coal hydrogasification
with co-production of electricity and synthetic natural gas. Int J Hydrogen Energy
2014;39(11):5793–803. http://dx.doi.org/10.1016/j.ijhydene.2014.01.155. ISSN:
03603199.
[38] Mccoy S, Rubin E. An engineering-economic model of pipeline transport of CO2
with application to carbon capture and storage. Int J Greenhouse Gas Control
2008;2(2):219–29. http://dx.doi.org/10.1016/S1750-5836(07)00119-3. ISSN:
17505836.
[39] Aspelund A, Tveit SP, Gundersen T. A liquefied energy chain for transport and
utilization of natural gas for power production with CO2 capture and storage – Part
3: The combined carrier and onshore storage. Appl Energy 2009;86(6):805–14.
http://dx.doi.org/10.1016/j.apenergy.2008.10.023. ISSN: 0306-2619.
[40] VTT Technical Research Centre of Finland. Toolkit: feasibility of power-to-gas
production; 2017.
[41] Statistics Finland. Population projection 2015 according to age and sex by area
2015–2040; 2016.
[42] The Finnish Energy Authority. Emission register; 2014.
[43] Holopainen A. Carbon sources for power-to-gas applications in the Finnish energy
system. MSc Thesis; 2015.
[44] Chemical Engineering Plant Cost Index. Chem Eng J 163 (1–2). ISSN: 1385-8947.
[45] Chemical Engineering Plant Cost Index. Chem Eng J 102 (3). ISSN: 1385-8947.
[46] Moore FT. Economies of scale: some statistical evidence. Quart J Econ
1959;73(2):232–45. http://dx.doi.org/10.2307/1883722. ISSN: 0033-5533.
[47] Chemical Engineering Plant Cost Index. Chem Eng J 143 (1–3). ISSN: 1385-8947.
[48] Götz M, Lefebvre J, Mörs F, McDaniel Koch A, Graf F, Bajohr S, Reimert R, Kolb T,
et al. Renewable power-to-gas: a technological and economic review. Renew Energy
2016;85:1371–90. http://dx.doi.org/10.1016/j.renene.2015.07.066. ISSN: 0960-
1481.
[49] Abanades J, Arias B, Lyngfelt A, Mattisson T, Wiley D, Li H, Ho M, Mangano E,
Brandani S, et al. Emerging CO2 capture systems. Int J Greenhouse Gas Control
2015;40:126–66. http://dx.doi.org/10.1016/j.ijggc.2015.04.018. ISSN: 17505836.